Allo-BMT has been an effective therapy for patients with myeloid malignancies due to a GVL effect. One approach designed to minimize GVHD side-effects is donor BM T cell depletion (TCD). Although TCD reduces GVHD, TCD can result in T-cell immune deficiency post-BMT. In Aim 1, we hypothesize that targeting the innate immune system (NK cells), which returns later post-BMT, will be the optimal approach to decrease AML recurrence after allo-BMT. Strategies are proposed to increase NK cells and block NK inhibitory receptors which will decrease AML recurrence by blocking the "off signal" delivered by AML cells to NK cells (Aim 1). In Aim 2, we hypothesize that the profound T-cell immunodeficiency post-BMT can be circumvented by strategies which protect the thymus against conditioning cells from conditioning regimen-induced injury. We will build upon this and determine whether DC vaccines will induce long-lasting memory cell responses in KGF treated recipients. We hypothesize that fusions of DCs and AML cells may be preferable to AML-lysate pulsed DCs due to superior loading of the MHC class I pathway via an endogenous rather than exogenous route. In another approach, we will induce DCs in vivo to present AML antigens by infusing AML-derived heat-shock proteins (hsps). Hsps associated with tumor peptides that are taken up by APCs can chaperone these peptides into MHC class I pathways and facilitate DCs maturation. We envision a sequential strategy in which thymus protective agents are given pre-BMT followed by approaches which stimulate the innate system early post-BMT and others that are directed toward stimulating the adaptive system late post-BMT.